Event Structure of Transitive Verb: A MARVS perspective

Module-Attribute Representation of Verbal Semantics (MARVS) is a theory of the representation of verbal semantics that is based on Mandarin Chinese data (Huang et al. 2000). In the MARVS theory, there are two different types of modules: Event Structure Modules and Role Modules. There are also two sets of attributes: Event-Internal Attributes and Role-Internal Attributes, which are linked to the Event Structure Module and the Role Module, respectively. In this study, we focus on four transitive verbs as chi1(eat), wan2(play), huan4(change) and shao1(burn) and explore their event structures by the MARVS theory.

💡 Research Summary

The paper applies the Module‑Attribute Representation of Verbal Semantics (MARVS), a theory originally developed on Mandarin data, to the analysis of four Chinese transitive verbs: chi1 (吃 “eat”), wan2 (玩 “play”), huan4 (换 “change”), and shao1 (烧 “burn”). MARVS distinguishes two kinds of modules—Event Structure Modules and Role Modules—and attaches two parallel sets of attributes, Event‑Internal Attributes and Role‑Internal Attributes, to them. The authors first outline the theoretical architecture of MARVS, describing the five canonical event structure types (Boundary, Process, State, Punctual, Stage) and the standard semantic roles (Agent, Theme, Goal, Source, Instrument, Beneficiary, etc.). They then map each of the four verbs onto this framework, showing how the combination of module type and attribute values captures the verb’s telicity, result state, and internal sub‑event composition.

For chi1 “eat”, the analysis identifies a three‑stage structure: an initial Boundary (food entering the mouth), a Process (chewing and swallowing), and a Result (the creation of a new physiological state, i.e., digestion). The Event‑Internal Attribute is telic, reflecting a clear endpoint, while the Role‑Internal Attributes involve an Agent (the eater) and a Theme (the food), with optional Instrument or Goal roles when context requires. Wan2 “play” is treated as an atelic Process‑only verb: the activity lacks a defined endpoint and does not generate a stable result state. Consequently, its Event‑Internal Attribute is atelic, and the Role‑Internal module expands to include not only Agent and Theme but also possible Instruments (toys, games) and Beneficiaries (other participants), illustrating the flexibility of MARVS in handling role‑rich, result‑poor events.

Huan4 “change” exhibits a more complex three‑phase pattern: a first Boundary marking the departure from the initial state, a Process denoting the transformation, and a second Boundary that establishes the new state. The Event‑Internal Attribute remains telic because the event is goal‑directed toward a specific outcome. Role‑Internal attributes involve Agent (the changer), Theme (the entity being changed), Goal (the new state), and Source (the prior state), demonstrating how MARVS can encode simultaneous role relations in a single event. Shao1 “burn” is modeled as a Punctual‑Result construction: a momentary Boundary (the ignition) followed by a sustained Result (the burned residue). The telic Event‑Internal Attribute reflects the creation of a new physical condition, and the Role‑Internal module includes Agent (the igniter), Theme (the object burned), Instrument (the fire source), and Goal (the transformation into ash or char).

Through systematic comparison, the authors argue that MARVS successfully differentiates between telic and atelic events, captures the presence or absence of result states, and accommodates both simple and composite event structures within a unified representational scheme. The study highlights the theory’s capacity to handle verbs with clear end‑states (chi1, shao1) as well as those where the outcome is vague or non‑existent (wan2), and shows that detailed role‑module specifications enrich the semantic description beyond what traditional lexical‑conceptual‑structure approaches provide.

The paper acknowledges several limitations. The empirical base is narrow, confined to four verbs, which restricts the generalizability of the findings. No statistical validation or corpus‑based frequency analysis is presented, and the treatment of more complex ditransitive or causative constructions is left for future work. The authors suggest extending the analysis to a larger set of transitive and intransitive verbs, incorporating corpus statistics, and possibly conducting psycholinguistic experiments to test the predictive power of MARVS attributes.

In conclusion, the research demonstrates that MARVS offers a robust, modular framework for dissecting the event semantics of Chinese transitive verbs. By linking event‑type modules with finely‑graded internal attributes and by elaborating role‑module configurations, the theory provides a nuanced account of telicity, resultivity, and role interaction, thereby contributing a valuable tool to the field of lexical semantics and to cross‑linguistic investigations of verb meaning.